ARTIFICIAL INTELLIGENCE AND CRIMINAL LIABILITY: A GLOBAL ANALYSIS OF THE RISKS OF ALGORITHMS IN THE CRIMINAL JUSTICE SYSTEM

Abstract

The integration of artificial intelligence (AI) into criminal justice systems has revolutionized decision-making processes but, at the same time, has raised profound legal and ethical issues. This article critically examines the various risks arising from the use of algorithmic systems in the judicial process, particularly with regard to criminal liability. Through a normative-comparative approach to the legal frameworks in the United States, Germany, China, and the European Union, this study identifies structural gaps in the regulation of automated decisions that can lead to harm or injustice. Findings indicate that conventional criminal liability doctrines are inadequate to address the challenges posed by the non-human and autonomous actions of AI systems. In response, this study proposes a hybrid liability framework that combines command responsibility and sociotechnical liability approaches into a new theoretical model aimed at achieving accountability in a digitized legal context. Furthermore, this study encourages international legal harmonization to ensure equal protection, as well as the establishment of global standards for transparency and oversight in the use of algorithms within criminal justice systems. Thus, this article contributes to the development of criminal law that is adaptive to digital realities and algorithmic governance.